Cable terminal simulation and state evaluation method and system in humid environment

Abstract

The invention relates to the technical field of cables, and particularly discloses a cable terminal simulation and state evaluation method and system in a humid environment, multiple types of sensor arrays are deployed, distributed temperature and humidity sensors are installed at key parts of a cable terminal, and aiming at the technical problems of one-sided single-dimensional monitoring, simulation model solidification and insufficient static weight adaptation, the cable terminal simulation and state evaluation method and system are provided. Cooperative acquisition of multiple parameters such as temperature and humidity, partial discharge and corrosion current is achieved through deployment of multiple types of sensor arrays, the nonlinear interaction relation between humidity permeation and electric field distortion is dynamically reflected in combination with a multi-physics field coupling simulation model, the one-sidedness of single parameter monitoring is solved, and the risk misjudgment rate is reduced by 40% or above; a dynamic water film thickness model is adopted to update interface parameters of the sealing ring in real time, a simulation result is updated per hour in combination with a finite element-boundary element hybrid algorithm, an electric field distortion simulation error is controlled within 12%, and the simulation precision is improved by more than one time compared with a fixed boundary condition.

Description

Technical Field

[0001] This invention relates to the field of cable technology, and in particular to a method and system for simulating and assessing the condition of cable terminals in humid environments. Background Technology

[0002] As a critical component of power transmission, the operational status assessment of cables requires a comprehensive evaluation of multiple parameters, including environmental humidity, partial discharge, and temperature. Current technologies typically employ single sensors to periodically monitor specific indicators, such as detecting partial discharge using high-frequency current methods or monitoring conductor heating using fiber optic thermometry. Mainstream methods primarily rely on threshold comparisons or empirical formulas for electrical parameter analysis; for example, using electric field strength thresholds to determine insulation defects or assessing insulation aging using dielectric loss angle data. Some improved solutions introduce finite element simulation technology to model electric field distribution, but these models are built based on standard environmental parameters and have fixed boundary conditions.

[0003] However, some shortcomings still need to be improved. First, the single-dimensional monitoring is one-sided, relying only on single parameters such as partial discharge and temperature, which cannot reflect the nonlinear interaction between humidity penetration and electric field distortion. The insulation degradation rate under high humidity is 3-5 times that in dry environment for the same discharge amount, which can easily lead to misjudgment of risks. Second, the simulation model is rigid, and the water film thickness at the interface between the sealing ring and the stress cone is fixed at 2μm, which cannot respond to humidity changes, and the simulation error of electric field distortion exceeds 25%. Third, the static weight adaptation is insufficient, and the parameter weight is difficult to adapt to sudden discharge or equipment aging. Fourth, the noise suppression is poor. Rainwater, electromagnetic interference and other interferences mask defects, and the filtering has a humidity-coupled noise suppression rate of less than 50%, and 30% of minor defects are misjudged as normal. Fifth, the aging assessment is lagging, and the moisture accumulation effect is not quantified, and the early warning of hidden defects is delayed by 3 cycles. Sixth, the feature correlation is missing. There is no mapping relationship between humidity and discharge, and defect identification fails when the signal-to-noise ratio is below 10dB. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for simulating and assessing the condition of cable terminals in a humid environment, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method and system for simulating and assessing the condition of cable terminals in a humid environment, comprising the following steps:

[0006] S1: Deploy multi-type sensor arrays, install distributed temperature and humidity sensors at key parts of cable terminals with sampling point spacing ≤3cm, ultra-high frequency partial discharge sensors (300MHz-1.5GHz), fiber optic temperature measurement units and corrosion current sensors, and integrate environmental sensors to monitor humidity, temperature, humidity change rate and salt spray concentration. All sensors upload data in real time through industrial Ethernet and use a timestamp encryption mechanism to ensure the traceability of original data.

[0007] S2: Performs spatiotemporal calibration and noise suppression for multi-source data. Time calibration is based on an optical fiber synchronous clock. A dynamic time bending algorithm is used to calculate the optimal matching path for partial discharge and environmental data, achieving a timing alignment error of ≤5ms. Spatial calibration is based on a three-dimensional coordinate model of the cable terminal. Delaunay triangulation is used to construct the topological relationship of monitoring points, and spatial attenuation factors are introduced to correct the data in the edge region. Noise processing uses a wavelet packet decomposition-independent component analysis joint algorithm. A dedicated filtering module is designed for humidity-coupled noise, which improves the signal-to-noise ratio by ≥15dB.

[0008] S3: Construct a multiphysics simulation model with dynamic boundary conditions. Based on real-time humidity data, calculate the interface parameters of the sealing ring using the water film thickness model h(φ)=0.12√φ+0.05×(dφ / dt). The three-dimensional model contains 12 fine structures such as conductors and insulating layers. Use the finite element-boundary element hybrid algorithm to solve the coupled field distribution of humidity diffusion, temperature field, electric field and metal corrosion. Update the simulation results every hour.

[0009] S4: Extract multi-dimensional risk assessment features. Time-domain features include 12 indicators such as partial discharge pulse amplitude and corrosion current change rate. Frequency-domain features obtain the energy proportion of 5 frequency bands through wavelet packet decomposition. Time-frequency domain features generate time-frequency matrix using short-time Fourier transform and extract features such as singular value entropy. Dimensionality reduction is achieved through the Relief-F algorithm, retaining the top 20 core features by weight.

[0010] S5: Establish a dynamic correlation mechanism between humidity and discharge. Construct a cross-correlation matrix between the two with a time window of 10 minutes. Introduce a time decay factor θ = e^(-t / 30) to weight and iterate the historical data. Extract the main eigenvalues ​​through matrix singular value decomposition to generate a dynamic coupling feature vector.

[0011] S6: Design an environment-aging coupling model, use partial least squares regression to calculate the correlation coefficient between dynamic coupling characteristics and salt spray concentration and humidity, determine the environmental factor weights (humidity 0.4, salt spray 0.3, temperature 0.3) through the analytic hierarchy process, introduce the equipment aging coefficient β = 1 + 0.03 × operating years, and construct the comprehensive coupling coefficient K = ω1 × c1 + ω2 × c2 + β × ω3 × c3;

[0012] S7: Construct an adaptive threshold adjustment system. Based on historical normal data, use kernel density estimation to generate an initial threshold range, and combine it with the real-time coupling coefficient K to dynamically correct it through a piecewise function: when K < 0.3, the threshold is lowered by 10%; when 0.3 ≤ K ≤ 0.7, the baseline value is maintained; when K > 0.7, it is increased by 15-25%; the threshold is corrected annually according to the aging coefficient.

[0013] S8: An improved deep learning evaluation model is adopted, with bidirectional LSTM as the main architecture. The input layer receives 20-dimensional dynamically coupled features, and the hidden layer has 3 layers (64-32-16 nodes). An attention mechanism is added to strengthen the weight of partial discharge features. The training adopts the Adam-particle swarm optimization algorithm with a learning rate of 0.001. The early stopping method (patience=15) is used to prevent overfitting, and the accuracy of 5-fold cross-validation is ≥96%.

[0014] S9: Design a four-level early warning assessment system: Level 1 (health index ≥ 85): No abnormal characteristics; Level 2 (70 ≤ health index < 85): Slight fluctuations; Level 3 (55 ≤ health index < 70): Monitoring needs to be strengthened; Level 4 (health index < 55): Emergency warning; When outputting, a feature contribution report is generated simultaneously to clarify the impact ratio of factors such as humidity penetration and partial discharge, establish a closed-loop optimization mechanism, and update the risk association rule library every quarter.

[0015] Furthermore, the generation of the standardized multi-source dataset in step S2 includes the following sub-steps:

[0016] S2.1: Collect real-time environmental data. Through a distributed temperature and humidity sensor array deployed on the surface of the cable terminal and the surrounding environment, with a sampling point spacing of ≤3cm, collect environmental humidity, temperature, humidity change rate and salt spray concentration data at different locations at a sampling frequency of 1Hz, and generate a real-time environmental dataset.

[0017] S2.2: Collect real-time electrical data, obtain the current fluctuation rate parameters, voltage deviation rate parameters, partial discharge pulse sequence (including amplitude, phase, frequency, and pulse width) of the cable body, and the dielectric loss angle of the insulation layer and the corrosion current of the metal components from the preset electrical monitoring terminal, and generate a real-time electrical dataset.

[0018] S2.3: Retrieve historical fault data, match the material type, service life, historical humidity exposure and fault repair records of the current cable terminal according to the preset historical fault database, and generate a historical fault mode feature set including micro-moisture concentration-aging correlation curve and corrosion rate-humidity relationship model;

[0019] S2.4: Perform data time-series alignment. The real-time environmental dataset, real-time electrical dataset, and historical fault mode feature set are aligned using an adaptive sliding window method. The window length is dynamically adjusted according to the humidity change rate. When the humidity change rate is ≥5% / h, the window length is shortened to 1 / 3 of the original length. When the humidity change rate is <2% / h, the window length is extended to twice the original length.

[0020] S2.5: Implement noise filtering processing. Use a joint algorithm of wavelet packet decomposition, singular value thresholding and independent component analysis to eliminate high-frequency noise interference on the aligned data. For partial discharge signals, an additional adaptive noise cancellation technique based on reference signals is used. Design a dedicated filtering module for humidity-coupled noise to achieve a humidity-coupled noise suppression rate of ≥85% and generate a standardized multi-source dataset.

[0021] Furthermore, the generation of dynamic finite element simulation data in step S3 includes the following sub-steps:

[0022] S3.1: Determine the dynamic boundary conditions. Based on the real-time humidity data, humidity change rate, and salt spray concentration in the standardized multi-source dataset, determine the interface water film thickness parameters of the cable terminal sealing ring and stress cone using the dynamic water film thickness model h(φ,dφ / dt,c)=0.12√φ+0.05×(dφ / dt)+0.03×c (φ is relative humidity, dφ / dt is humidity change rate, and c is salt spray concentration). The water film thickness range is 0.5-15μm.

[0023] S3.2: Construct a three-dimensional simulation model. Based on the determined dynamic boundary conditions, build a three-dimensional spatial numerical simulation model of the cable terminal, which includes fine structures such as conductors and insulation layers.

[0024] S3.3: Perform multi-physics coupling solution. In the three-dimensional space numerical simulation model, the micro-water concentration distribution in the cable insulation layer is solved based on the humidity diffusion equation (introducing the humidity-dependent diffusion coefficient D(φ)=D0·e^(k·φ), where k is a material constant with a value range of 0.02-0.05). Combined with the temperature field equation (coupled Joule heat, dielectric loss heat and natural convection heat dissipation), the electric field distribution equation (considering the water film conductivity σ=10^(-4)~10^(-2)S / m and the surface conductivity change caused by salt spray corrosion) and the metal corrosion kinetic equation, coupled iterative calculation is performed to generate the dielectric loss angle parameter, conductivity distribution parameter, micro-water concentration gradient parameter and metal corrosion depth parameter of the insulation material.

[0025] S3.4: Verify and generate data. Using a preset Monte Carlo sampling method, simulate the dynamic boundary conditions with ±20% random perturbation to verify the 95% confidence intervals of the dielectric loss angle parameter, conductivity distribution parameter, micro-water concentration gradient parameter, and metal corrosion depth parameter. Generate dynamic finite element simulation data with error range markings to ensure that the electric field distortion simulation error is ≤12%.

[0026] Furthermore, the generation of the multi-dimensional risk assessment feature set in step S4 includes the following sub-steps:

[0027] S4.1: Extract key simulation features. Extract the peak value of electric field distortion in the humidity penetration area, electric field distortion rate (ratio of distorted area to total area), temperature rise gradient of insulation layer, micro water accumulation (micro water content per unit volume) and metal corrosion depth from the dynamic finite element simulation data to generate a set of key simulation features.

[0028] S4.2: Extract measured features. Extract the duration, amplitude change rate, phase distribution entropy, high frequency component ratio (frequency > 1MHz), and corrosion current change rate of the partial discharge pulse from the standardized multi-source dataset to generate a measured partial discharge and corrosion feature set.

[0029] S4.3: Establish an association rule base, map the simulation key feature set with the measured partial discharge and corrosion feature set through fuzzy logic association rules, establish a nonlinear relationship model of humidity fluctuation-micro water concentration-electric field distortion-partial discharge-metal corrosion, and generate a risk association rule base containing more than 50 association rules. The rule base contains dynamic correction coefficients for discharge feature and corrosion feature thresholds under different humidity ranges and salt spray concentrations.

[0030] S4.4: Generate a multi-dimensional risk assessment feature set. Based on the risk association rule base, combined with the historical failure mode feature set, equipment aging coefficient (β = 1 + 0.03 × operating years + 0.01 × cumulative humidity exposure) and environmental severity coefficient (γ = 1 + 0.02 × salt spray concentration), predict the potential failure types (such as stress cone creepage, insulation breakdown, sheath corrosion perforation, etc.), occurrence probability and remaining life assessment value of the current cable terminal, and generate the multi-dimensional risk assessment feature set.

[0031] Furthermore, the dynamic weight allocation algorithm in step S6 includes the following sub-steps:

[0032] S6.1: Determine the initial weights. Based on the preset analytic hierarchy process and combined with expert experience and historical fault data, determine the initial weight allocation: humidity parameter 0.25, electric field distortion parameter 0.25, partial discharge parameter 0.25, temperature parameter 0.1, corrosion parameter 0.15, and generate the initial weight allocation table.

[0033] S6.2: Dynamically adjust weights based on the humidity fluctuation rate parameter, salt spray concentration, and equipment aging coefficient in the risk association rule base: when the humidity change rate is ≥5% / h, the humidity weight is increased by 30%; when the salt spray concentration is ≥0.05mg / m³, the weight is increased by 30%. 3 The weight of corrosion parameters is increased by 25%; when the equipment has been in operation for more than 10 years, the weight of partial discharge is increased by 20%.

[0034] S6.3: Set priority weights. If the frequency of partial discharge in the standardized multi-source dataset exceeds a preset critical threshold (the critical threshold is dynamically set according to humidity and salt spray concentration; the threshold decreases by 20% for every 10% increase in humidity and by 0.01 mg / m³ for every 0.01 mg / m³ increase in salt spray concentration), the priority weight will be determined. 3 If the threshold is reduced by 15%, the priority of the partial discharge weight parameter will be raised to the highest level, temporarily increased to 2.5 times the initial value, and a dynamic weight allocation result will be generated.

[0035] Furthermore, the execution of the model recalibration command in step S8 includes the following sub-steps:

[0036] S8.1: Trigger recalibration signal. When the prediction error of the cable health index exceeds the preset error threshold (8%) for three consecutive evaluation cycles, or the current humidity deviates from the model calibration humidity by ≥25%, or the salt spray concentration exceeds the historical maximum value, the recalibration trigger signal is triggered.

[0037] S8.2: Correct the model parameters. Based on the real-time humidity data, humidity change rate, salt spray concentration and recent partial discharge and corrosion characteristic trends in the standardized multi-source dataset, correct the boundary condition water film thickness parameters, material dielectric constant temperature coefficient and corrosion rate constant of the three-dimensional spatial numerical simulation model.

[0038] S8.3: Update the association rule base, update the non-linear relationship model in the risk association rule base, introduce the latest failure cases and laboratory simulation data within the last 6 months, and retrain the association rules;

[0039] S8.4: Optimize weight allocation, reallocate the initial values ​​of the weight parameters in the dynamic weight allocation algorithm, use the Bayesian optimization algorithm to optimize the weight allocation, and generate an optimized comprehensive evaluation model;

[0040] S8.5: Fine-tune the evaluation model. Use an online incremental learning algorithm to fine-tune the optimized comprehensive evaluation model by accumulating 800-1200 sets of new data. Update the training set through a sliding window and retain the latest 6 months of data to ensure that the prediction error of the model after recalibration is ≤6%.

[0041] Furthermore, the method also includes the following steps:

[0042] S7.1: Generate dielectric aging assessment parameters. Perform multivariate nonlinear regression analysis on the dielectric loss angle parameter, cumulative micro-water concentration, partial discharge pulse amplitude change rate, and corrosion current change rate to generate dielectric aging assessment parameters A=tanδ×(1+0.08×C_w)×(1+0.1×ΔU / U0)×(1+0.15×ΔI / I0) (C_w is the cumulative micro-water concentration, ΔU / U0 is the partial discharge amplitude change rate, and ΔI / I0 is the corrosion current change rate).

[0043] S7.2: Correct the probability of failure. Based on the dielectric aging assessment parameters, equipment service life, and cumulative humidity exposure, dynamically correct the predicted probability of failure in the risk association rule base. When A≥0.3, the predicted probability of failure is multiplied by a correction factor of 1.5; when A≥0.5, the predicted probability of failure is multiplied by a correction factor of 2.5, generating an updated multi-dimensional risk assessment feature set.

[0044] S7.3: Adjust the evaluation and sampling frequency. When the dielectric aging degree evaluation parameter exceeds the warning threshold, automatically shorten the evaluation cycle from 1 hour to 10 minutes and increase the sampling frequency of partial discharge and corrosion current to 5 times the original frequency.

[0045] Furthermore, S9.1: Establish a historical trend database, establish a historical trend database of cable health index, store health index data for the past 3 years, use triple exponential smoothing combined with seasonal index adjustment to fit the long-term trend curve, and dynamically adjust the smoothing coefficient according to seasonal humidity characteristics, reduce the smoothing coefficient to 0.2 during the summer high humidity period, and increase the smoothing coefficient to 0.6 during the winter dry period.

[0046] S9.2: Trigger forced recalibration. If the deviation of the current cable health index from the historical trend exceeds the preset deviation threshold (18%), or the health index decrease rate is ≥4% / cycle for 5 consecutive cycles, or the insulation state sub-index is below 60, then the forced recalibration command of the three-dimensional spatial numerical simulation model is triggered.

[0047] S9.3: Generate an assessment report, feed back the recalibrated cable health index to the comprehensive assessment model, update the early warning signal trigger logic, set four levels of early warning: Level 1 early warning (health index ≥ 85), Level 2 early warning (70 ≤ health index < 85), Level 3 early warning (55 ≤ health index < 70), and Level 4 early warning (health index < 55), and generate an assessment report that includes fault location, risk level, and recommended maintenance plan.

[0048] 9. A simulation and condition assessment system for cable terminals in a humid environment, the system comprising:

[0049] The data acquisition module is used to acquire environmental data, electrical monitoring data, and historical fault record data of the cable terminal, and generate a raw multi-source dataset. The environmental data includes real-time humidity data. The raw multi-source dataset is then time-series aligned and noise filtered to generate a standardized multi-source dataset.

[0050] The simulation modeling module is used to construct a three-dimensional spatial numerical simulation model of the cable terminal based on the real-time humidity data in the standardized multi-source dataset, and to perform coupled solution of humidity diffusion, temperature field and electric field distribution based on the three-dimensional spatial numerical simulation model to generate dynamic finite element simulation data.

[0051] The risk assessment module is used to integrate the dynamic finite element simulation data with the electrical monitoring data in the standardized multi-source dataset, extract the electric field distortion parameters, partial discharge pulse frequency parameters and insulation layer temperature rise gradient parameters in the humidity penetration area, and generate a multi-dimensional risk assessment feature set.

[0052] The dynamic assessment module is used to input the multi-dimensional risk assessment feature set into a preset comprehensive assessment model, calculate the cable health index through a dynamic weight allocation algorithm, and trigger an early warning signal or model recalibration command based on the cable health index.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] Firstly, this invention addresses the technical problems of one-dimensional monitoring being insufficient, fixed simulation models, and inadequate static weight adaptation. It achieves coordinated acquisition of multiple parameters such as temperature, humidity, partial discharge, and corrosion current through the deployment of multi-type sensor arrays. Combined with a multi-physics field coupled simulation model, it dynamically reflects the nonlinear interaction between humidity penetration and electric field distortion, solving the one-sidedness of single-parameter monitoring and reducing the risk misjudgment rate by more than 40%. A dynamic water film thickness model is used to update the sealing ring interface parameters in real time, and a finite element-boundary element hybrid algorithm updates the simulation results hourly, controlling the electric field distortion simulation error to within 12%, more than doubling the accuracy compared to simulations under fixed boundary conditions. Furthermore, a dynamic weight allocation algorithm adjusts parameter weights in real time based on humidity change rate and equipment aging coefficient, improving the assessment accuracy by 25% in sudden discharge or equipment aging scenarios, significantly enhancing the assessment model's adaptability to complex working conditions.

[0055] Secondly, this invention addresses the technical problems of poor noise suppression, delayed aging assessment, and missing feature correlations by employing a wavelet packet decomposition-independent component analysis joint algorithm and a dedicated humidity-coupled noise filtering module. This improves the signal-to-noise ratio by more than 15 dB, achieves a humidity-coupled noise suppression rate of over 85%, and reduces the false positive rate for minor defects to below 5%. By quantifying the moisture accumulation effect through dielectric aging degree assessment parameters, early warnings can be triggered at the nascent stage of hidden defects, detecting risks three cycles earlier than traditional methods and effectively avoiding sudden failures. Furthermore, by constructing a humidity-discharge dynamic correlation mechanism and a multi-dimensional feature mapping relationship, the defect identification accuracy remains above 90% even in complex environments with a signal-to-noise ratio below 10 dB, significantly improving the stability of defect identification in low signal-to-noise ratio scenarios. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating a method for simulating and assessing the condition of cable terminals in a humid environment, according to an embodiment of the present invention.

[0057] Figure 2 This is a schematic diagram of the structure of a cable terminal simulation and condition assessment system in a humid environment according to an embodiment of the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Please see Figure 1 In this embodiment of the invention, a method for simulating and assessing the condition of cable terminals in a humid environment includes the following steps:

[0060] S1: Deploy multi-type sensor arrays, install distributed temperature and humidity sensors at key parts of cable terminals with sampling point spacing ≤3cm, ultra-high frequency partial discharge sensors (300MHz-1.5GHz), fiber optic temperature measurement units and corrosion current sensors, and integrate environmental sensors to monitor humidity, temperature, humidity change rate and salt spray concentration. All sensors upload data in real time through industrial Ethernet and use a timestamp encryption mechanism to ensure the traceability of original data.

[0061] S2: Performs spatiotemporal calibration and noise suppression for multi-source data. Time calibration is based on an optical fiber synchronous clock. A dynamic time bending algorithm is used to calculate the optimal matching path for partial discharge and environmental data, achieving a timing alignment error of ≤5ms. Spatial calibration is based on a three-dimensional coordinate model of the cable terminal. Delaunay triangulation is used to construct the topological relationship of monitoring points, and spatial attenuation factors are introduced to correct the data in the edge region. Noise processing uses a wavelet packet decomposition-independent component analysis joint algorithm. A dedicated filtering module is designed for humidity-coupled noise, which improves the signal-to-noise ratio by ≥15dB.

[0062] S3: Construct a multiphysics simulation model with dynamic boundary conditions. Based on real-time humidity data, calculate the interface parameters of the sealing ring using the water film thickness model h(φ)=0.12√φ+0.05×(dφ / dt). The three-dimensional model contains 12 fine structures such as conductors and insulating layers. Use the finite element-boundary element hybrid algorithm to solve the coupled field distribution of humidity diffusion, temperature field, electric field and metal corrosion. Update the simulation results every hour.

[0063] S4: Extract multi-dimensional risk assessment features. Time-domain features include 12 indicators such as partial discharge pulse amplitude and corrosion current change rate. Frequency-domain features obtain the energy proportion of 5 frequency bands through wavelet packet decomposition. Time-frequency domain features generate time-frequency matrix using short-time Fourier transform and extract features such as singular value entropy. Dimensionality reduction is achieved through the Relief-F algorithm, retaining the top 20 core features by weight.

[0064] S5: Establish a dynamic correlation mechanism between humidity and discharge. Construct a cross-correlation matrix between the two with a time window of 10 minutes. Introduce a time decay factor θ = e^(-t / 30) to weight and iterate the historical data. Extract the main eigenvalues ​​through matrix singular value decomposition to generate a dynamic coupling feature vector.

[0065] S6: Design an environment-aging coupling model, use partial least squares regression to calculate the correlation coefficient between dynamic coupling characteristics and salt spray concentration and humidity, determine the environmental factor weights (humidity 0.4, salt spray 0.3, temperature 0.3) through the analytic hierarchy process, introduce the equipment aging coefficient β = 1 + 0.03 × operating years, and construct the comprehensive coupling coefficient K = ω1 × c1 + ω2 × c2 + β × ω3 × c3;

[0066] S7: Construct an adaptive threshold adjustment system. Based on historical normal data, use kernel density estimation to generate an initial threshold range, and combine it with the real-time coupling coefficient K to dynamically correct it through a piecewise function: when K < 0.3, the threshold is lowered by 10%; when 0.3 ≤ K ≤ 0.7, the baseline value is maintained; when K > 0.7, it is increased by 15-25%; the threshold is corrected annually according to the aging coefficient.

[0067] S8: An improved deep learning evaluation model is adopted, with bidirectional LSTM as the main architecture. The input layer receives 20-dimensional dynamically coupled features, and the hidden layer has 3 layers (64-32-16 nodes). An attention mechanism is added to strengthen the weight of partial discharge features. The training adopts the Adam-particle swarm optimization algorithm with a learning rate of 0.001. The early stopping method (patience=15) is used to prevent overfitting, and the accuracy of 5-fold cross-validation is ≥96%.

[0068] S9: Design a four-level early warning assessment system: Level 1 (health index ≥ 85): No abnormal characteristics; Level 2 (70 ≤ health index < 85): Slight fluctuations; Level 3 (55 ≤ health index < 70): Monitoring needs to be strengthened; Level 4 (health index < 55): Emergency warning; When outputting, a feature contribution report is generated simultaneously to clarify the impact ratio of factors such as humidity penetration and partial discharge, establish a closed-loop optimization mechanism, and update the risk association rule library every quarter.

[0069] The step S2 of generating a standardized multi-source dataset includes the following sub-steps:

[0070] S2.1: Collect real-time environmental data. Through a distributed temperature and humidity sensor array deployed on the surface of the cable terminal and the surrounding environment, with a sampling point spacing of ≤3cm, collect environmental humidity, temperature, humidity change rate and salt spray concentration data at different locations at a sampling frequency of 1Hz, and generate a real-time environmental dataset.

[0071] S2.2: Collect real-time electrical data, obtain the current fluctuation rate parameters, voltage deviation rate parameters, partial discharge pulse sequence (including amplitude, phase, frequency, and pulse width) of the cable body, and the dielectric loss angle of the insulation layer and the corrosion current of the metal components from the preset electrical monitoring terminal, and generate a real-time electrical dataset.

[0072] S2.3: Retrieve historical fault data, match the material type, service life, historical humidity exposure and fault repair records of the current cable terminal according to the preset historical fault database, and generate a historical fault mode feature set including micro-moisture concentration-aging correlation curve and corrosion rate-humidity relationship model;

[0073] S2.4: Perform data time-series alignment. The real-time environmental dataset, real-time electrical dataset, and historical fault mode feature set are aligned using an adaptive sliding window method. The window length is dynamically adjusted according to the humidity change rate. When the humidity change rate is ≥5% / h, the window length is shortened to 1 / 3 of the original length. When the humidity change rate is <2% / h, the window length is extended to twice the original length.

[0074] S2.5: Implement noise filtering processing. Use a joint algorithm of wavelet packet decomposition, singular value thresholding and independent component analysis to eliminate high-frequency noise interference on the aligned data. For partial discharge signals, an additional adaptive noise cancellation technique based on reference signals is used. Design a dedicated filtering module for humidity-coupled noise to achieve a humidity-coupled noise suppression rate of ≥85% and generate a standardized multi-source dataset.

[0075] The generation of dynamic finite element simulation data in step S3 includes the following sub-steps:

[0076] S3.1: Determine the dynamic boundary conditions. Based on the real-time humidity data, humidity change rate, and salt spray concentration in the standardized multi-source dataset, determine the interface water film thickness parameters of the cable terminal sealing ring and stress cone using the dynamic water film thickness model h(φ,dφ / dt,c)=0.12√φ+0.05×(dφ / dt)+0.03×c (φ is relative humidity, dφ / dt is humidity change rate, and c is salt spray concentration). The water film thickness range is 0.5-15μm.

[0077] S3.2: Construct a three-dimensional simulation model. Based on the determined dynamic boundary conditions, build a three-dimensional spatial numerical simulation model of the cable terminal, which includes fine structures such as conductors and insulation layers.

[0078] S3.3: Perform multi-physics coupling solution. In the three-dimensional space numerical simulation model, the micro-water concentration distribution in the cable insulation layer is solved based on the humidity diffusion equation (introducing the humidity-dependent diffusion coefficient D(φ)=D0·e^(k·φ), where k is a material constant with a value range of 0.02-0.05). Combined with the temperature field equation (coupled Joule heat, dielectric loss heat and natural convection heat dissipation), the electric field distribution equation (considering the water film conductivity σ=10^(-4)~10^(-2)S / m and the surface conductivity change caused by salt spray corrosion) and the metal corrosion kinetic equation, coupled iterative calculation is performed to generate the dielectric loss angle parameter, conductivity distribution parameter, micro-water concentration gradient parameter and metal corrosion depth parameter of the insulation material.

[0079] S3.4: Verify and generate data. Using a preset Monte Carlo sampling method, simulate the dynamic boundary conditions with ±20% random perturbation to verify the 95% confidence intervals of the dielectric loss angle parameter, conductivity distribution parameter, micro-water concentration gradient parameter, and metal corrosion depth parameter. Generate dynamic finite element simulation data with error range markings to ensure that the electric field distortion simulation error is ≤12%.

[0080] The step S4 of generating a multi-dimensional risk assessment feature set includes the following sub-steps:

[0081] S4.1: Extract key simulation features. Extract the peak value of electric field distortion in the humidity penetration area, electric field distortion rate (ratio of distorted area to total area), temperature rise gradient of insulation layer, micro water accumulation (micro water content per unit volume) and metal corrosion depth from the dynamic finite element simulation data to generate a set of key simulation features.

[0082] S4.2: Extract measured features. Extract the duration, amplitude change rate, phase distribution entropy, high frequency component ratio (frequency > 1MHz), and corrosion current change rate of the partial discharge pulse from the standardized multi-source dataset to generate a measured partial discharge and corrosion feature set.

[0083] S4.3: Establish an association rule base, map the simulation key feature set with the measured partial discharge and corrosion feature set through fuzzy logic association rules, establish a nonlinear relationship model of humidity fluctuation-micro water concentration-electric field distortion-partial discharge-metal corrosion, and generate a risk association rule base containing more than 50 association rules. The rule base contains dynamic correction coefficients for discharge feature and corrosion feature thresholds under different humidity ranges and salt spray concentrations.

[0084] S4.4: Generate a multi-dimensional risk assessment feature set. Based on the risk association rule base, combined with the historical failure mode feature set, equipment aging coefficient (β = 1 + 0.03 × operating years + 0.01 × cumulative humidity exposure) and environmental severity coefficient (γ = 1 + 0.02 × salt spray concentration), predict the potential failure types (such as stress cone creepage, insulation breakdown, sheath corrosion perforation, etc.), occurrence probability and remaining life assessment value of the current cable terminal, and generate the multi-dimensional risk assessment feature set.

[0085] The dynamic weight allocation algorithm in step S6 includes the following sub-steps:

[0086] S6.1: Determine the initial weights. Based on the preset analytic hierarchy process and combined with expert experience and historical fault data, determine the initial weight allocation: humidity parameter 0.25, electric field distortion parameter 0.25, partial discharge parameter 0.25, temperature parameter 0.1, corrosion parameter 0.15, and generate the initial weight allocation table.

[0087] S6.2: Dynamically adjust weights based on the humidity fluctuation rate parameter, salt spray concentration, and equipment aging coefficient in the risk association rule base: when the humidity change rate is ≥5% / h, the humidity weight is increased by 30%; when the salt spray concentration is ≥0.05mg / m³, the weight is increased by 30%. 3 The weight of corrosion parameters is increased by 25%; when the equipment has been in operation for more than 10 years, the weight of partial discharge is increased by 20%.

[0088] S6.3: Set priority weights. If the frequency of partial discharge in the standardized multi-source dataset exceeds a preset critical threshold (the critical threshold is dynamically set according to humidity and salt spray concentration; the threshold decreases by 20% for every 10% increase in humidity and by 0.01 mg / m³ for every 0.01 mg / m³ increase in salt spray concentration), the priority weight will be determined. 3 If the threshold is reduced by 15%, the priority of the partial discharge weight parameter will be raised to the highest level, temporarily increased to 2.5 times the initial value, and a dynamic weight allocation result will be generated.

[0089] The execution of the model recalibration command in step S8 includes the following sub-steps:

[0090] S8.1: Trigger recalibration signal. When the prediction error of the cable health index exceeds the preset error threshold (8%) for three consecutive evaluation cycles, or the current humidity deviates from the model calibration humidity by ≥25%, or the salt spray concentration exceeds the historical maximum value, the recalibration trigger signal is triggered.

[0091] S8.2: Correct the model parameters. Based on the real-time humidity data, humidity change rate, salt spray concentration and recent partial discharge and corrosion characteristic trends in the standardized multi-source dataset, correct the boundary condition water film thickness parameters, material dielectric constant temperature coefficient and corrosion rate constant of the three-dimensional spatial numerical simulation model.

[0092] S8.3: Update the association rule base, update the non-linear relationship model in the risk association rule base, introduce the latest failure cases and laboratory simulation data within the last 6 months, and retrain the association rules;

[0093] S8.4: Optimize weight allocation, reallocate the initial values ​​of the weight parameters in the dynamic weight allocation algorithm, use the Bayesian optimization algorithm to optimize the weight allocation, and generate an optimized comprehensive evaluation model;

[0094] S8.5: Fine-tune the evaluation model. Use an online incremental learning algorithm to fine-tune the optimized comprehensive evaluation model by accumulating 800-1200 sets of new data. Update the training set through a sliding window and retain the latest 6 months of data to ensure that the prediction error of the model after recalibration is ≤6%.

[0095] The method further includes the following steps:

[0096] S7.1: Generate dielectric aging assessment parameters. Perform multivariate nonlinear regression analysis on the dielectric loss angle parameter, cumulative micro-water concentration, partial discharge pulse amplitude change rate, and corrosion current change rate to generate dielectric aging assessment parameters A=tanδ×(1+0.08×C_w)×(1+0.1×ΔU / U0)×(1+0.15×ΔI / I0) (C_w is the cumulative micro-water concentration, ΔU / U0 is the partial discharge amplitude change rate, and ΔI / I0 is the corrosion current change rate).

[0097] S7.2: Correct the probability of failure. Based on the dielectric aging assessment parameters, equipment service life, and cumulative humidity exposure, dynamically correct the predicted probability of failure in the risk association rule base. When A≥0.3, the predicted probability of failure is multiplied by a correction factor of 1.5; when A≥0.5, the predicted probability of failure is multiplied by a correction factor of 2.5, generating an updated multi-dimensional risk assessment feature set.

[0098] S7.3: Adjust the evaluation and sampling frequency. When the dielectric aging degree evaluation parameter exceeds the warning threshold, automatically shorten the evaluation cycle from 1 hour to 10 minutes and increase the sampling frequency of partial discharge and corrosion current to 5 times the original frequency.

[0099] S9.1: Establish a historical trend database, store the historical trend of cable health index, and fit the long-term trend curve using the triple exponential smoothing method combined with seasonal index adjustment. The smoothing coefficient is dynamically adjusted according to the seasonal humidity characteristics. The smoothing coefficient is reduced to 0.2 during the summer high humidity period and increased to 0.6 during the winter dry period.

[0100] S9.2: Trigger forced recalibration. If the deviation of the current cable health index from the historical trend exceeds the preset deviation threshold (18%), or the health index decrease rate is ≥4% / cycle for 5 consecutive cycles, or the insulation state sub-index is below 60, then the forced recalibration command of the three-dimensional spatial numerical simulation model is triggered.

[0101] S9.3: Generate an assessment report, feed back the recalibrated cable health index to the comprehensive assessment model, update the early warning signal trigger logic, set four levels of early warning: Level 1 early warning (health index ≥ 85), Level 2 early warning (70 ≤ health index < 85), Level 3 early warning (55 ≤ health index < 70), and Level 4 early warning (health index < 55), and generate an assessment report that includes fault location, risk level, and recommended maintenance plan.

[0102] Please see Figure 2 A simulation and condition assessment system for cable terminals in humid environments, the system comprising:

[0103] The data acquisition module is used to acquire environmental data, electrical monitoring data, and historical fault record data of the cable terminal, and generate a raw multi-source dataset. The environmental data includes real-time humidity data. The raw multi-source dataset is then time-series aligned and noise filtered to generate a standardized multi-source dataset.

[0104] The simulation modeling module is used to construct a three-dimensional spatial numerical simulation model of the cable terminal based on the real-time humidity data in the standardized multi-source dataset, and to perform coupled solution of humidity diffusion, temperature field and electric field distribution based on the three-dimensional spatial numerical simulation model to generate dynamic finite element simulation data.

[0105] The risk assessment module is used to integrate the dynamic finite element simulation data with the electrical monitoring data in the standardized multi-source dataset, extract the electric field distortion parameters, partial discharge pulse frequency parameters and insulation layer temperature rise gradient parameters in the humidity penetration area, and generate a multi-dimensional risk assessment feature set.

[0106] The dynamic assessment module is used to input the multi-dimensional risk assessment feature set into a preset comprehensive assessment model, calculate the cable health index through a dynamic weight allocation algorithm, and trigger an early warning signal or model recalibration command based on the cable health index.

[0107] The operating principle and advantages of this invention are as follows: Environmental parameters of cable terminals, such as temperature, humidity, and salt spray concentration, as well as electrical parameters, such as partial discharge and corrosion current, and historical fault data, are collected in real time by deploying multi-type sensor arrays. Standardized multi-source datasets are generated after spatiotemporal calibration and noise suppression. A multi-physics simulation model with dynamic boundary conditions is constructed based on real-time humidity data. The coupled field distribution is solved using a finite element-boundary element hybrid algorithm to generate dynamic finite element simulation data. Multi-dimensional risk features are extracted by fusing simulation data and measured data. A dynamic weight allocation algorithm and an improved deep learning model are used to calculate the cable health index. A four-level early warning system is implemented by combining this with an adaptive threshold adjustment system. The risk association rule base is continuously updated through a closed-loop optimization mechanism, forming a full-process state management system from data collection, processing, simulation, evaluation to early warning. First, by using multi-source data collaborative monitoring and dynamic coupling analysis, it breaks through the limitations of single-dimensional monitoring, accurately capturing the nonlinear relationship between humidity and electric field distortion, thus improving the risk assessment accuracy to over 96%. Second, by employing dynamic boundary condition simulation and real-time model recalibration, it solves the problem of traditional model rigidity, controlling the electric field distortion simulation error to within 12%, significantly improving simulation accuracy. Third, by combining dynamic weight allocation and adaptive threshold adjustment, it can adapt to equipment aging and environmental changes, maintaining assessment stability even in sudden discharge or high humidity scenarios. Fourth, through a dedicated noise filtering module and multi-domain feature extraction, the humidity coupling noise suppression rate reaches over 85%, and the minor defect identification rate is improved by 30%. Fifth, by quantifying the moisture accumulation effect and dielectric aging parameters, it can detect hidden defects three cycles in advance, avoiding sudden failures. Sixth, the system has closed-loop optimization capabilities, updating the rule base quarterly, and can adapt to the complex operating environment of cable terminals for a long time, reducing maintenance costs and extending equipment life.

[0108] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any method of indirect connection is applicable to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0109] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for simulating and assessing the condition of cable terminals in a humid environment, characterized in that, Includes the following steps: S1: Deploy multi-type sensor arrays, install distributed temperature and humidity sensors at key parts of cable terminals with sampling point spacing ≤3cm, ultra-high frequency partial discharge sensors (300MHz-1.5GHz), fiber optic temperature measurement units and corrosion current sensors, and integrate environmental sensors to monitor humidity, temperature, humidity change rate and salt spray concentration. All sensors upload data in real time through industrial Ethernet and use a timestamp encryption mechanism to ensure the traceability of original data. S2: Performs spatiotemporal calibration and noise suppression for multi-source data. Time calibration is based on an optical fiber synchronous clock. A dynamic time bending algorithm is used to calculate the optimal matching path for partial discharge and environmental data, achieving a timing alignment error of ≤5ms. Spatial calibration is based on a three-dimensional coordinate model of the cable terminal. Delaunay triangulation is used to construct the topological relationship of monitoring points, and spatial attenuation factors are introduced to correct the data in the edge region. Noise processing uses a wavelet packet decomposition-independent component analysis joint algorithm. A dedicated filtering module is designed for humidity-coupled noise, which improves the signal-to-noise ratio by ≥15dB. S3: Construct a multiphysics simulation model with dynamic boundary conditions. Based on real-time humidity data, calculate the interface parameters of the sealing ring using the water film thickness model h(φ)=0.12√φ+0.05×(dφ / dt). The three-dimensional model contains 12 fine structures such as conductors and insulating layers. Use the finite element-boundary element hybrid algorithm to solve the coupled field distribution of humidity diffusion, temperature field, electric field and metal corrosion. Update the simulation results every hour. S4: Extract multi-dimensional risk assessment features, including 12 indicators such as partial discharge pulse amplitude and corrosion current change rate in the time domain; Frequency domain features are obtained by wavelet packet decomposition to obtain the energy proportion of 5 frequency bands; time-frequency domain features are generated by short-time Fourier transform to extract features such as singular value entropy; dimensionality reduction is performed by the Relief-F algorithm to retain the top 20 core features by weight. S5: Establish a dynamic correlation mechanism between humidity and discharge. Construct a cross-correlation matrix between the two with a time window of 10 minutes. Introduce a time decay factor θ = e^(-t / 30) to weight and iterate the historical data. Extract the main eigenvalues ​​through matrix singular value decomposition to generate a dynamic coupling feature vector. S6: Design an environment-aging coupling model, use partial least squares regression to calculate the correlation coefficient between dynamic coupling characteristics and salt spray concentration and humidity, determine the environmental factor weights (humidity 0.4, salt spray 0.3, temperature 0.3) through the analytic hierarchy process, introduce the equipment aging coefficient β = 1 + 0.03 × operating years, and construct the comprehensive coupling coefficient K = ω1 × c1 + ω2 × c2 + β × ω3 × c3; S7: Construct an adaptive threshold adjustment system. Based on historical normal data, use kernel density estimation to generate an initial threshold range, and combine it with the real-time coupling coefficient K to dynamically correct it through a piecewise function: when K < 0.3, the threshold is lowered by 10%; when 0.3 ≤ K ≤ 0.7, the baseline value is maintained; when K > 0.7, it is increased by 15-25%; the threshold is corrected annually according to the aging coefficient. S8: An improved deep learning evaluation model is adopted, with bidirectional LSTM as the main architecture. The input layer receives 20-dimensional dynamically coupled features, and the hidden layer has 3 layers (64-32-16 nodes). An attention mechanism is added to strengthen the weight of partial discharge features. Training employs the Adam-particle swarm optimization algorithm with a learning rate of 0.

001. Overfitting is prevented by early stopping (patience = 15), and the accuracy of 5-fold cross-validation is ≥96%. S9: Design a four-level early warning assessment system, Level 1 (health index ≥ 85): No abnormal characteristics; Level 2 (70 ≤ Health Index < 85): Slight fluctuations; Level 3 (55 ≤ Health Index < 70): Increased monitoring required; Level 4 (Health Index < 55): Emergency warning. During output, a feature contribution report is generated simultaneously to clarify the impact ratio of factors such as humidity penetration and partial discharge, establish a closed-loop optimization mechanism, and update the risk association rule library every quarter.

2. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 1, characterized in that, The step S2 of generating a standardized multi-source dataset includes the following sub-steps: S2.1: Collect real-time environmental data. Through a distributed temperature and humidity sensor array deployed on the surface of the cable terminal and the surrounding environment, with a sampling point spacing of ≤3cm, collect environmental humidity, temperature, humidity change rate and salt spray concentration data at different locations at a sampling frequency of 1Hz, and generate a real-time environmental dataset. S2.2: Collect real-time electrical data, obtain the current fluctuation rate parameters, voltage deviation rate parameters, partial discharge pulse sequence (including amplitude, phase, frequency, and pulse width) of the cable body, and the dielectric loss angle of the insulation layer and the corrosion current of the metal components from the preset electrical monitoring terminal, and generate a real-time electrical dataset. S2.3: Retrieve historical fault data, match the material type, service life, historical humidity exposure and fault repair records of the current cable terminal according to the preset historical fault database, and generate a historical fault mode feature set including micro-moisture concentration-aging correlation curve and corrosion rate-humidity relationship model; S2.4: Perform data time-series alignment. The real-time environmental dataset, real-time electrical dataset, and historical fault mode feature set are aligned using an adaptive sliding window method. The window length is dynamically adjusted according to the humidity change rate. When the humidity change rate is ≥5% / h, the window length is shortened to 1 / 3 of the original length. When the humidity change rate is <2% / h, the window length is extended to twice the original length. S2.5: Implement noise filtering processing. Use a joint algorithm of wavelet packet decomposition, singular value thresholding and independent component analysis to eliminate high-frequency noise interference on the aligned data. For partial discharge signals, an additional adaptive noise cancellation technique based on reference signals is used. Design a dedicated filtering module for humidity-coupled noise to achieve a humidity-coupled noise suppression rate of ≥85% and generate a standardized multi-source dataset.

3. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 1, characterized in that, The generation of dynamic finite element simulation data in step S3 includes the following sub-steps: S3.1: Determine the dynamic boundary conditions. Based on the real-time humidity data, humidity change rate, and salt spray concentration in the standardized multi-source dataset, determine the interface water film thickness parameters of the cable terminal sealing ring and stress cone using the dynamic water film thickness model h(φ,dφ / dt,c)=0.12√φ+0.05×(dφ / dt)+0.03×c (φ is relative humidity, dφ / dt is humidity change rate, and c is salt spray concentration). The water film thickness range is 0.5-15μm. S3.2: Construct a three-dimensional simulation model. Based on the determined dynamic boundary conditions, build a three-dimensional spatial numerical simulation model of the cable terminal, which includes fine structures such as conductors and insulation layers. S3.3: Perform multi-physics coupling solution. In the three-dimensional space numerical simulation model, the micro-water concentration distribution in the cable insulation layer is solved based on the humidity diffusion equation (introducing the humidity-dependent diffusion coefficient D(φ)=D0·e^(k·φ), where k is a material constant with a value range of 0.02-0.05). Combined with the temperature field equation (coupled Joule heat, dielectric loss heat and natural convection heat dissipation), the electric field distribution equation (considering the water film conductivity σ=10^(-4)~10^(-2)S / m and the surface conductivity change caused by salt spray corrosion) and the metal corrosion kinetic equation, coupled iterative calculation is performed to generate the dielectric loss angle parameter, conductivity distribution parameter, micro-water concentration gradient parameter and metal corrosion depth parameter of the insulation material. S3.4: Verify and generate data. Using a preset Monte Carlo sampling method, simulate the dynamic boundary conditions with ±20% random perturbation to verify the 95% confidence intervals of the dielectric loss angle parameter, conductivity distribution parameter, micro-water concentration gradient parameter, and metal corrosion depth parameter. Generate dynamic finite element simulation data with error range markings to ensure that the electric field distortion simulation error is ≤12%.

4. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 3, characterized in that, The step S4 of generating a multi-dimensional risk assessment feature set includes the following sub-steps: S4.1: Extract key simulation features. Extract the peak value of electric field distortion in the humidity penetration area, electric field distortion rate (ratio of distorted area to total area), temperature rise gradient of insulation layer, micro water accumulation (micro water content per unit volume) and metal corrosion depth from the dynamic finite element simulation data to generate a set of key simulation features. S4.2: Extract measured features. Extract the duration, amplitude change rate, phase distribution entropy, high frequency component ratio (frequency > 1MHz), and corrosion current change rate of the partial discharge pulse from the standardized multi-source dataset to generate a measured partial discharge and corrosion feature set. S4.3: Establish an association rule base, map the simulation key feature set with the measured partial discharge and corrosion feature set through fuzzy logic association rules, establish a nonlinear relationship model of humidity fluctuation-micro water concentration-electric field distortion-partial discharge-metal corrosion, and generate a risk association rule base containing more than 50 association rules. The rule base contains dynamic correction coefficients for discharge feature and corrosion feature thresholds under different humidity ranges and salt spray concentrations. S4.4: Generate a multi-dimensional risk assessment feature set. Based on the risk association rule base, combined with the historical failure mode feature set, equipment aging coefficient (β = 1 + 0.03 × operating years + 0.01 × cumulative humidity exposure) and environmental severity coefficient (γ = 1 + 0.02 × salt spray concentration), predict the potential failure types (such as stress cone creepage, insulation breakdown, sheath corrosion perforation, etc.), occurrence probability and remaining life assessment value of the current cable terminal, and generate the multi-dimensional risk assessment feature set.

5. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 4, characterized in that, The dynamic weight allocation algorithm in step S6 includes the following sub-steps: S6.1: Determine the initial weights. Based on the preset analytic hierarchy process and combined with expert experience and historical fault data, determine the initial weight allocation: humidity parameter 0.25, electric field distortion parameter 0.25, partial discharge parameter 0.25, temperature parameter 0.1, corrosion parameter 0.15, and generate the initial weight allocation table. S6.2: Dynamically adjust weights based on the humidity fluctuation rate parameter, salt spray concentration, and equipment aging coefficient in the risk association rule base: when the humidity change rate is ≥5% / h, the humidity weight is increased by 30%; when the salt spray concentration is ≥0.05mg / m³, the weight is increased by 30%. 3 The weight of corrosion parameters is increased by 25%; when the equipment has been in operation for more than 10 years, the weight of partial discharge is increased by 20%. S6.3: Set priority weights. If the frequency of partial discharge in the standardized multi-source dataset exceeds a preset critical threshold (the critical threshold is dynamically set according to humidity and salt spray concentration; the threshold decreases by 20% for every 10% increase in humidity and by 0.01 mg / m³ for every 0.01 mg / m³ increase in salt spray concentration), the priority weight will be determined. 3 If the threshold is reduced by 15%, the priority of the partial discharge weight parameter will be raised to the highest level, temporarily increased to 2.5 times the initial value, and a dynamic weight allocation result will be generated.

6. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 5, characterized in that, The execution of the model recalibration command in step S8 includes the following sub-steps: S8.1: Trigger recalibration signal. When the prediction error of the cable health index exceeds the preset error threshold (8%) for three consecutive evaluation cycles, or the current humidity deviates from the model calibration humidity by ≥25%, or the salt spray concentration exceeds the historical maximum value, the recalibration trigger signal is triggered. S8.2: Correct the model parameters. Based on the real-time humidity data, humidity change rate, salt spray concentration and recent partial discharge and corrosion characteristic trends in the standardized multi-source dataset, correct the boundary condition water film thickness parameters, material dielectric constant temperature coefficient and corrosion rate constant of the three-dimensional spatial numerical simulation model. S8.3: Update the association rule base, update the non-linear relationship model in the risk association rule base, introduce the latest failure cases and laboratory simulation data within the last 6 months, and retrain the association rules; S8.4: Optimize weight allocation, reallocate the initial values ​​of the weight parameters in the dynamic weight allocation algorithm, use the Bayesian optimization algorithm to optimize the weight allocation, and generate an optimized comprehensive evaluation model; S8.5: Fine-tune the evaluation model. Use an online incremental learning algorithm to fine-tune the optimized comprehensive evaluation model by accumulating 800-1200 sets of new data. Update the training set through a sliding window and retain the latest 6 months of data to ensure that the prediction error of the model after recalibration is ≤6%.

7. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 4, characterized in that, The method further includes the following steps: S7.1: Generate dielectric aging assessment parameters. Perform multivariate nonlinear regression analysis on the dielectric loss angle parameter, cumulative micro-water concentration, partial discharge pulse amplitude change rate, and corrosion current change rate to generate dielectric aging assessment parameters A=tanδ×(1+0.08×C_w)×(1+0.1×ΔU / U0)×(1+0.15×ΔI / I0) (C_w is the cumulative micro-water concentration, ΔU / U0 is the partial discharge amplitude change rate, and ΔI / I0 is the corrosion current change rate). S7.2: Correct the probability of failure. Based on the dielectric aging assessment parameters, equipment service life, and cumulative humidity exposure, dynamically correct the predicted value of the probability of failure in the risk association rule base. When A≥0.3, the predicted value of the probability of failure is multiplied by a correction factor of 1.

5. When A≥0.5, the predicted failure probability is multiplied by a correction factor of 2.5 to generate an updated multi-dimensional risk assessment feature set; S7.3: Adjust the evaluation and sampling frequency. When the dielectric aging degree evaluation parameter exceeds the warning threshold, automatically shorten the evaluation cycle from 1 hour to 10 minutes and increase the sampling frequency of partial discharge and corrosion current to 5 times the original frequency.

8. The method for simulating and assessing the condition of cable terminals in a humid environment according to claim 1, characterized in that, The method further includes the following steps: S9.1: Establish a historical trend database, store the historical trend of cable health index, and fit the long-term trend curve using the triple exponential smoothing method combined with seasonal index adjustment. The smoothing coefficient is dynamically adjusted according to the seasonal humidity characteristics. The smoothing coefficient is reduced to 0.2 during the summer high humidity period and increased to 0.6 during the winter dry period. S9.2: Trigger forced recalibration. If the deviation of the current cable health index from the historical trend exceeds the preset deviation threshold (18%), or the health index decrease rate is ≥4% / cycle for 5 consecutive cycles, or the insulation state sub-index is below 60, then the forced recalibration command of the three-dimensional spatial numerical simulation model is triggered. S9.3: Generate an assessment report, feed back the recalibrated cable health index to the comprehensive assessment model, update the early warning signal trigger logic, set four levels of early warning: Level 1 early warning (health index ≥ 85), Level 2 early warning (70 ≤ health index < 85), Level 3 early warning (55 ≤ health index < 70), and Level 4 early warning (health index < 55), and generate an assessment report that includes fault location, risk level, and recommended maintenance plan.

9. A simulation and condition assessment system for cable terminals in a humid environment, the system comprising: The data acquisition module is used to acquire environmental data, electrical monitoring data, and historical fault record data of the cable terminal, and generate a raw multi-source dataset. The environmental data includes real-time humidity data. The raw multi-source dataset is then time-series aligned and noise filtered to generate a standardized multi-source dataset. The simulation modeling module is used to construct a three-dimensional spatial numerical simulation model of the cable terminal based on the real-time humidity data in the standardized multi-source dataset, and to perform coupled solution of humidity diffusion, temperature field and electric field distribution based on the three-dimensional spatial numerical simulation model to generate dynamic finite element simulation data. The risk assessment module is used to integrate the dynamic finite element simulation data with the electrical monitoring data in the standardized multi-source dataset, extract the electric field distortion parameters, partial discharge pulse frequency parameters and insulation layer temperature rise gradient parameters in the humidity penetration area, and generate a multi-dimensional risk assessment feature set. The dynamic assessment module is used to input the multi-dimensional risk assessment feature set into a preset comprehensive assessment model, calculate the cable health index through a dynamic weight allocation algorithm, and trigger an early warning signal or model recalibration command based on the cable health index.

Images